Of course it’s an opinion: the point is that it’s neither persuasive nor internally inconsistent. They haven’t given any reason to believe they have enough domain knowledge to compare the two authoritatively. It’s also inconsistent to criticize OOXML for being difficult to implement without extra knowledge and then to criticize a truly open spec for being detailed enough to implement without extra – the entire HTML5 process was intended to reduce the number of cases where people were relying on things which required implementers to know how a specific engine like IE worked.